Test Equipment Anonymous

Took a chance on a pig-in-a-poke on the Bay of Evil and snagged an 8664A 0.1~3000 MHz signal generator about two weeks ago.
Incoming inspection is of course a requirement:


The listing showed it connected to a spec an and putting out a nice 3 GHz signal, so it was at least somewhat operational. Upon arrival it took a long time to go through a calibration and boot, and of course booted with an unknown gibberish result code along with hardware failure and calibration error indications. The the result code not having either a leading + or - sign indicated it wasn't a legitimate failure detected by the system during bootup, so this combined with the long self cal led to checking the RAM backup battery; a 3.6 V NiCd. It was stone dead, and couldn't be blasted into response by hitting it with a higher charging voltage (after, of course, disconnecting it from the circuit), so a replacement was ordered from Battery Mart.


The batteries arrived today and I installed the replacement on the processor board:


To my very pleasant surprise tonight, after two boot cycles and a reset, it began booting cleanly without any errors or messages, and appears to be functional. The stickers and residue cleaned up nicely, and the instrument looks to be in very good nick. I think it's the newest piece of 'vintage' test gear that I have at approximately a quarter century old. I need to wring it out thoroughly, but it looks good at first brush!


Final inspection by Widlar:


-Pat

You lucky thing. Looks like a very nice piece of kit, if rather large. My E4420B (2GHz) generator is a little bit smaller, being only 3U high, but makes up for it by being very deep.

In my limited experience, NiMH has a less tolerant charging regime than NiCd. It might be worth checking the charging regime and comparing that to the NiMH's spec.

tggzzz wrote: ↑Tue Apr 07, 2026 8:27 am In my limited experience, NiMH has a less tolerant charging regime than NiCd. It might be worth checking the charging regime and comparing that to the NiMH's spec.

Definitely. Plenty of information on the WWW. IIRC NiCd is still approved of for emergency lighting. The better NiMH chargers monitor for Delta-V, the slight drop in voltage which occurs when the cell is charged, and some also monitor temperature changes. The cells have to be charged with enough current to show these changes, which as I recall is at or over 1/3 C. Charging cells in series seems a bad idea, because they all have slightly different requirements. Of course, if you have a battery pack of cells, it would add a lot to the cost to arrange for them to be charged individually. I'd imagine the designers looked into all this and the NiMH battery didn't cause many problems with their charging regime.

I believe NiMH don't tolerate being stored fully discharged for months, which has probably been the case here. With batteries of NiMH cells, if they are flattened, there will always be a weakest cell driven backwards, which is damaging. I'd guess the RAM backup current is too much for a lithium cell, commonly used to keep alive the BIOS settings in PCs.

Anyway, the NiMH battery was replaced, and as long as the instrument is powered on every few months, it should last for years.

Nice find by the way.

IIRC NiMH trickle charging current is much lower than NiCd.

For batteries, I wonder if any of the Advanced Linear Devices EPAD MOSFETs might be useful; they certainly are for balancing supercaps. Basic principle: they electrically adjust the Vth in 0.1mV increments, so it is easy to have matched pairs/arrays.

http://www.aldinc.com/pdf/AN1108.pdf
http://www.aldinc.com/ald_refmatls.php

Not always: take the Tek1502, which uses a PUJT to stuff pulses into the 9-cell NiCd battery. Mean current might be OK, but the crest factor is ridiculous. Even Tek had to issue a note saying "don't charge continuously for more than 16 hours" (and recharge once/month, because even when switched off there is a 1mA battery drain). it is a wonder Patriot missile batteries were ever operational!

As for Li? Too many variant chemistries and manufacturing corners being cut.

Apparently some Li C/D/etc cells have a built in BMS IC, which ought to address the former issue.

Yeah, I need to open it up and dig further into the charging circuitry - the original battery was a NiCd, and I searched on the part number (N-50SB3), and that's the one that came up from Battery Mart as a replacement.


I foolishly didn't dig further, figured it was a NiCd and didn't notice until after installing it that it is in fact NiMH. Of course to complicate things, this instrument is from HP's later "no schematics included" generation, so this will be a PITA. (And of course I also have since found that Newark has the 3.6 V NiCd pack for something like $8-9 instead of the $30 I paid for what I stupidly thought was an exact replacement.)


Oh well, at least the generator works - in the grand scheme of things the battery is a minor issue. I may just order the NiCd from Newark and chalk it up as a lesson learned.

-Pat

EC8010 wrote: ↑Tue Apr 07, 2026 8:17 am You lucky thing. Looks like a very nice piece of kit, if rather large. My E4420B (2GHz) generator is a little bit smaller, being only 3U high, but makes up for it by being very deep.

Thanks! Yeah, it's definitely a big honker - about 81 lbs/37 kg and 24"/610 mm deep. I felt bad for the poor UPS guy; it came when I wasn't home so I couldn't help carry it.

-Pat

Cubdriver wrote: ↑Tue Apr 07, 2026 5:16 pm
I foolishly didn't dig further, figured it was a NiCd and didn't notice until after installing it that it is in fact NiMH. Of course to complicate things, this instrument is from HP's later "no schematics included" generation, so this will be a PITA. (And of course I also have since found that Newark has the 3.6 V NiCd pack for something like $8-9 instead of the $30 I paid for what I stupidly thought was an exact replacement.)


Oh well, at least the generator works - in the grand scheme of things the battery is a minor issue. I may just order the NiCd from Newark and chalk it up as a lesson learned.

-Pat

You could bung in a super capacitor. I haven't messed with them for 20 odd years, and that was a 0.5F I codged on a Z80 board in place of the recommended battery, but I have an idea they've come on since then. Those came from N R Bardwell, now sadly defunct, but at the time a supplier of mainly good but iffy parts.

Supercaps have a 2.7V rating, so you would need a couple in series. I believe such things can be purchased. I don't know how they balance the voltages; maybe there are ALD MOSFETs internally.

Don't forget that as a cell discharges, the voltage stays "constant" until it is exhausted. Contrariwise, a capacitor's voltage falls "linearly" over time.

tggzzz wrote: ↑Tue Apr 07, 2026 1:23 pm Not always: take the Tek1502, which uses a PUJT to stuff pulses into the 9-cell NiCd battery. Mean current might be OK, but the crest factor is ridiculous. Even Tek had to issue a note saying "don't charge continuously for more than 16 hours" (and recharge once/month, because even when switched off there is a 1mA battery drain). it is a wonder Patriot missile batteries were ever operational!

As for Li? Too many variant chemistries and manufacturing corners being cut.

Apparently some Li C/D/etc cells have a built in BMS IC, which ought to address the former issue.

I was specifically referring to memory backup applications. Perhaps I should have been more clear.

The problem with the built-in BMS is that it usually contains a very low RDSon MOSFET which disconnects the cell to protect it from over- and under-charging. Not ideal for memory backup.

Zenith wrote: ↑Tue Apr 07, 2026 9:29 pm

Cubdriver wrote: ↑Tue Apr 07, 2026 5:16 pm
I foolishly didn't dig further, figured it was a NiCd and didn't notice until after installing it that it is in fact NiMH. Of course to complicate things, this instrument is from HP's later "no schematics included" generation, so this will be a PITA. (And of course I also have since found that Newark has the 3.6 V NiCd pack for something like $8-9 instead of the $30 I paid for what I stupidly thought was an exact replacement.)


Oh well, at least the generator works - in the grand scheme of things the battery is a minor issue. I may just order the NiCd from Newark and chalk it up as a lesson learned.

-Pat

You could bung in a super capacitor. I haven't messed with them for 20 odd years, and that was a 0.5F I codged on a Z80 board in place of the recommended battery, but I have an idea they've come on since then. Those came from N R Bardwell, now sadly defunct, but at the time a supplier of mainly good but iffy parts.

Rechargeable cells/batteries are used when supercaps don't have the storage capacity. Which is to say, old school supercaps. Modern ones are quite different, and I wouldn't have one anywhere near the internals of an important instrument, given the short-circuit current they are capable of delivering.

Mine was definitely old school and was used to keep alive an SDRAM. The SDRAM would stay alive for several minutes with nothing connected anyway, so its demands were minuscule.

I'm sure you are right about modern super caps.

https://www.youtube.com/shorts/X6Y0-2Y6zEw